interpretation of a negative nonlinear Duffing stiffness (-x^3) in limit-cycle models

[Sina Mehdizadeh]

In modeling biologic rhythmic actions as limit-cycles, it has been reported that a negative Duffing term (i.e. -x^3) represents decreasing of variability near reversal points (or softening spring). The question is what type of "variability" we mean here?spatial?temporal?or any other type?
How could it be proved or visualized based on Duffing equations or any other method that variability decreases near reversal points?